Mamba
PulseAugur coverage of Mamba — every cluster mentioning Mamba across labs, papers, and developer communities, ranked by signal.
- used by Samba 95%
- developed by Samba 95%
- instance of State Space Models 90%
- used by lidar 90%
- used by electroencephalography 90%
- developed U-Net 90%
- uses CNN 80%
- used by magnetic resonance imaging 80%
- competes with attention 80%
- competes with CNN 70%
- used by State Space Models 70%
- used by alphaXiv 70%
19 day(s) with sentiment data
-
State-Space Models: From S4 to Mamba Reviewed
This paper provides a comprehensive review of Structured State Space Models (SSMs), tracing their evolution from the initial S4 architecture to more advanced models like Mamba and Mamba-2. It analyzes key design dimensi…
-
New AraSSM model offers efficient Arabic language processing
Researchers have introduced AraSSM, a new bidirectional Mamba encoder specifically designed for Arabic masked language modeling. This model aims to overcome the quadratic scaling limitations of traditional Transformer e…
-
KV Cache Emerges as LLM Bottleneck, Driving Attention Variant Innovations
The KV cache, a critical component in autoregressive decoding for LLMs, is identified as the primary bottleneck for frontier models in 2026. Its size grows linearly with context length and batch size, making it the domi…
-
New Ignition Index metric measures global workspace dynamics in language models
Researchers have introduced the Ignition Index (I), a new metric designed to quantify global workspace dynamics within language models. This scalar metric operationalizes predictions from Global Workspace Theory (GWT) b…
-
New frameworks enhance mask transformers and adapt State Space Models for missing data
Researchers have developed iFAN, a training framework designed to enhance mask transformers by aligning query ranking with mask quality and improving intermediate prediction distillation. This method addresses mismatche…
-
TSM-Pose framework enhances object pose estimation with topology and semantics
Researchers have introduced TSM-Pose, a novel framework designed to improve category-level object pose estimation. This method utilizes a Topology Extractor to capture global structural representations from point clouds…
-
Hierarchical Memory Mamba enhances long-sequence modeling
Researchers have introduced Hierarchical Memory Mamba (HMM), a novel architecture designed to enhance the long-sequence modeling capabilities of recurrent linear attention models like Mamba. By incorporating a hierarchi…
-
Mamba-based knowledge distillation enhances LiDAR 3D object detection
Researchers have developed a new knowledge distillation framework to improve the efficiency of 3D object detection using LiDAR sensors. This method transfers object-level voxel representations from a powerful teacher mo…
-
New Hear to See method advances audio-visual instance segmentation
Researchers have developed a new method called Hear to See (H2S) to improve audio-visual instance segmentation. This technique addresses the challenges of matching overlapping acoustic events with visual instances and h…
-
Muon optimizer shows token efficiency gains on Mamba-2 130M
A new research paper explores the effectiveness of the Muon optimizer, which uses a Newton-Schulz iteration for steepest descent under the spectral norm, on state-space models. The study compares Muon with AdamW on the …
-
MambaTS framework offers improved time series forecasting without self-attention
Researchers have developed MambaTS, a new framework for long-term time series forecasting that utilizes a selective state space model as its backbone. Unlike traditional Transformers, MambaTS avoids the quadratic comple…
-
Mamba Policy offers efficient 3D manipulation with reduced parameters
Researchers have developed a new AI policy model called Mamba Policy, which significantly reduces computational requirements for 3D manipulation tasks. This model utilizes a hybrid approach combining Mamba and Attention…
-
New T-PMambaSR framework enhances image super-resolution with Mamba and attention
Researchers have introduced T-PMambaSR, a new lightweight framework for image super-resolution that combines window-based self-attention with Progressive Mamba. This approach aims to capture global receptive fields effi…
-
USP-Mamba enhances hyperspectral image super-resolution with spectral and structural prompting
Researchers have developed USP-Mamba, a novel framework for hyperspectral image super-resolution that enhances Mamba-based models. This new approach addresses limitations in existing models by incorporating unmixing-der…
-
Loop-Mamba framework advances old photo restoration with novel memory techniques
Researchers have introduced Loop-Mamba, a novel framework designed for restoring old photographs. This system utilizes a loop-based state-space approach to progressively refine image restoration states through iterative…
-
New MSCM-net model combines CNNs and Mamba for hyperspectral image classification
Researchers have developed MSCM-net, a novel hyperspectral image classification model that combines multi-scale convolutional neural networks (CNNs) with Mamba blocks. This architecture aims to improve classification pe…
-
BladeYOLO framework enhances wind turbine defect detection with limited data
Researchers have developed BladeYOLO, a new framework designed to improve the detection of defects on wind turbine blades, particularly in scenarios with limited annotated data. The system integrates a Vision Transforme…
-
Satellite image change detection methods compared for text-based queries · 2 sources tracked
Researchers have evaluated eight different methods for combining before-and-after satellite images to efficiently answer text-based queries about changes. The study compared attention, Mamba, and Temporal Bottleneck Fus…
-
New SPFM-Net framework targets invisible watermarks with Mamba architecture
Researchers have developed SPFM-Net, a novel framework designed to attack invisible watermarks in images. This system utilizes a semantic-prior-guided and frequency-constrained Mamba architecture to effectively remove w…
-
Rad-JEPA 3D: New framework enhances 3D CT scan analysis
Researchers have introduced Rad-JEPA 3D, a new self-supervised learning framework designed for 3D medical image analysis, specifically for computed tomography (CT) scans. This model utilizes a hybrid H-Mamba encoder tha…